Founder Mode

Founder Mode. An essay by Paul Graham, September 2024.

Advice that backfired

At a Y Combinator event, Airbnb's Brian Chesky gave a talk many founders called the best they'd heard. His theme: the usual advice for growing companies is wrong.

While Airbnb was growing, he got the standard, well-meant advice: hire good people, then step back. He took it, and it was a disaster.

So he found his own way, partly by studying Steve Jobs's Apple, and it's working. In the room, many top founders said the same advice had hurt them too.

Founder mode, manager mode

Why the wrong advice? It was meant for a hired manager running someone else's company. Founders can do things managers can't, and holding back feels wrong.

In other words, companies can be run in two ways. Call them founder mode and manager mode. Nearly everyone assumed a growing startup had to switch to the second.

How do we know founder mode exists? From founders' dismay in manager mode, and their success once they broke out. Yet no book describes it.

Manager mode: black boxes

Manager mode works like modular software. Each branch of the org chart is a black box: the CEO hands it a goal, and never looks inside.

Reaching in to fix the details? That's micromanaging, which managers are taught to avoid. Just hire good people and trust them. It sounds lovely.

But founder after founder finds it often means, as Graham puts it, “hire professional fakers and let them drive the company into the ground.”

Gaslit from both sides

Founders said they felt gaslit, made to doubt their own eyes. From both sides: by advisers urging them to manage, then by their own people once they did.

Usually, when everyone says you're wrong, you are. It's a good rule. But Graham thinks founders have hit a rare exception.

Why? Investors who've never founded a company have no way of knowing how a founder should run one.

And senior executives as a group, he says, include some world-class liars. The tactful version: they're very good at managing up.

Skipping levels

We can already see one difference. In manager mode, the CEO deals only with direct reports. Founder mode will break that rule.

Talking to people further down is now so rare it has a name: skip-level meetings. They'd become routine, opening up countless arrangements.

Steve Jobs did a version of this. Once a year he took the 100 people he thought mattered most on a retreat, wherever they sat on the org chart.

It could make a huge company feel like a startup. Jobs kept holding them, so they likely worked, yet no one else seems to. Good idea? Nobody knows yet.

Borders that move

None of this means doing it all yourself. What works with 20 people can't work with 2,000. Some things must be handed off.

Where does each manager's freedom end? The line will differ by company, crisp or blurry, and move outward as a manager earns trust.

That makes founder mode harder to pin down than manager mode's one tidy rule. But it works better, as founders feeling their way toward it already show.

He expects misuse too: founders using it as an excuse not to delegate, and non-founders playing founder. Manager mode at least caps the harm a bad CEO can do.

Against a headwind

Graham expects some founders will turn out to have been doing it all along, more or less. People just called them eccentric, or worse.

Strangely, knowing so little is encouraging. Founders have achieved a lot already, with bad advice blowing in their faces the whole way.

Now imagine founders who get the right advice: how to run a company like Steve Jobs, not like John Sculley.

In short

Most advice on scaling teaches manager mode: how to run someone else's company.

Manager mode treats the org chart as black boxes. For founders, it often backfires.

Founder mode reaches past direct reports, but still delegates as trust grows.

It's barely understood yet. That's good news: founders can do even better.

The original is short, and better than this. Read it: paulgraham.com/foundermode.html